top of page
Search
  • Adi Shalom

"Less is More" - The Usage Rate Problem

Updated: Oct 15, 2021

You already have ads integrated in your game, and that’s great.

When most publishers are trying to increase their ad revenue it usually comes down to two main solutions:

1. Improving the CPM

2. Increasing the number of impressions per DAU


The equation seems simple enough.

Ad Revenue = (Impressions amount/1000) * CPM

The number of impressions is determined by the amount of Daily Engaged Users (DEU) and the average amount of ads per daily engaged user (Usage Rate).

Impressions per DAU = DEU * UR


From an integration standpoint, most publishers with active ads in the games have reached a point that they feel they know which users should see ads (depends on the game, genre, monetization strategy, segmentation capabilities, etc.).

So, their focus is now to increase the Usage Rate and thus increasing the overall number of impressions to make higher ad revenue.


Leaving aside potential concerns of game economy, potential abuse by users, user experience and risks to In-App Purchases if this is not configured properly, there are additional effects that might not lead to more ad revenue as well.


The Problem

Having more impressions does not necessarily lead to higher ad revenue.

This statement sounds strange, since it seems logical to assume that if you sell more (of any commodity), you should increase your sales.

However, the reason is for this statement is split into two:


1. “Impressions depth”

Most advertisers in gaming apps are other gaming apps, who buy users with a CPI model or deeper funnel events.

As a result, the greater the IPM (Installs Per 1000 impressions) is, the greater the CPM is for the publisher’s impressions.

When publishers increase their UR, the IPM will drop, leading to lower CPM.






If the UR was not very high to begin with, usually the CPM drop will be compensated by the increase of number of impressions, which leads to higher revenue.

But this is not always true.

In some cases, the CPM drop will be stronger than the impressions amount increase, which leads to revenue loss.


How can the CPM drop more than the increase of impressions?


First – the SDK networks algorithms start to react.

One of the common practices for UA who buy through SDK ad networks is to cut the sub-sites (publisher apps) that lower their campaigns overall CPM score.

So, some advertisers will stop advertising in your apps because of the IPM drop.

The advertisers obviously are unaware of the recent configuration changes done by publishers, and usually take a few days to react and to stop advertising in your apps.

As a result, the CPM drop (or fill rate drop) will not always be visible immediately after increasing the UR on the publisher side.

A common mistake by publishers who AB test such changes in UR, is to look at their results after a short period of time (which does not take into consideration the CPM drop that has yet to come) and making wrong decisions.


This effect of advertisers blocking a publisher app is usually irreversible as well, as most advertisers do not unblock publisher apps unless some specific reason or a new information that comes to light.


Second – the advertisers start to react.

There are many factors that cause the network’s algorithms to react the way they do.

As a disclaimer, I have never worked for any ad network, or claim to fully understand all factors that are taken into consideration.

However, the network’s basic agenda is clear.

When advertisers buy with a CPI model, and publishers are paid in a CPM model, the main variable that the network needs to predict to make a profit or to avoid losses is the IPM.

Networks need to buy impressions from publishers with a CPM bid before any install has occurred. When advertisers pay the network on a CPI basis, without an install, the network is not getting paid for the impressions they bought and are exposed to losses.

To minimize this risk, the network’s algorithms should be more cautious and to lower their bids when the facing rapid negative changes in IPM.


An increase in UR immediately impacts IPM and thus impact the network’s prediction models of the IPM leading to lower CPM or fill rate.

Since it is wise for networks to be more cautious when facing rapid negative changes, the algorithms might “overreact”, and lower CPM more than the actual performance has changed.


2. The oranges metaphor

Let’s put aside the topic at hand for a second and imagine yourself buying a crate of oranges every week.

You have been buying a crate with 100 perfect oranges for weeks for $100.

You think that these perfect oranges worth the price you pay and continue to do so.

One day the person who have been selling you these oranges for a while, offers you to buy an additional crate with 20 perfect oranges and 80 rotten oranges for $10 more.


If we are all logic driven, then we understand that we can buy more perfect oranges with a 50% discount, while throwing the rotten oranges away, and we would agree to buy the second crate.


However, over time, we might look at the deal a bit differently, and think that paying $110 for oranges that 40% are rotten is not such an amazing deal.

We might even stop buying any oranges from this seller at some point, because of this perspective caused by me judging the 2-crate deal as a whole.


Now let’s get back to our topic, and assume that the oranges are impressions, and the price is CPM.

By selling more low-quality impressions (rotten oranges), we might actually give the appearance that our overall quality of impressions is poor, even if we are actually able to deliver more installs to our advertisers.

Advertisers on the other side of the equation will see negative changes in their KPIs (purchases rate, IPM, etc.) which can lead them to lower bids, or block the publisher’s app all together.


Solutions

Well, there isn’t necessarily a problem to begin with if you understand the logic behind these chain reactions, and act accordingly from the start.


Gradual changes

Most of the problems described here can be solved by making gradual changes.

When my favorite chocolate bar suddenly had a small notice on it stating that its size is now 5% smaller, but the price stayed the same, I still bought it.

Eventually it shrunk over time more and more, but the price stayed the same, and I kept buying it.

If at some point the manufacturer would have written on the wrapping that the bar is now 20% smaller, I might have bought less or switched to some other snack (just for spite).

Similarly, advertisers are not expected to make drastic changes when quality stays more or less the same as it was before.


Compensate

If you do need to make quick changes, try to compensate for the changes you make.

There are ways to benefit our advertisers while making negative changes to keep them.

Temporarily under-optimizing specific price points in your waterfall might help your advertisers to see positive changes to contradict the changes in UR, and you can gradually change your price points back to an optimized state later on.


Inform your partners

You can also reach out to your networks account managers, asking them to stay alert for changes, and to notify their demand side as well to avoid advertisers making drastic changes. This is especially wise if the network knows that their traffic is heavily reliant on specific advertisers buying from your app at that time.


Conclusion

There are different ways to increase your ad revenue wisely.

Increasing the impressions per DAU can be the right move, but it’s not always as simple as it sounds.

There isn’t a rule of thumb when making changes in the ads configuration in your app and information is scarce, so try to think about the full impact of changes you make and how it impacts the partners you work with and the advertisers who buy your traffic.

1 view0 comments

Comentários


Post: Blog2_Post
bottom of page